Article Cite This: J. Am. Chem. Soc. 2017, 139, 15363-15370
pubs.acs.org/JACS
M2(m‑dobdc) (M = Mn, Fe, Co, Ni) Metal−Organic Frameworks as Highly Selective, High-Capacity Adsorbents for Olefin/Paraffin Separations Jonathan E. Bachman,†,∥ Matthew T. Kapelewski,‡,∥ Douglas A. Reed,‡ Miguel I. Gonzalez,‡ and Jeffrey R. Long*,†,‡,§ †
Department of Chemical & Biomolecular Engineering, and ‡Department of Chemistry, University of California, Berkeley, California 94720, United States § Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States S Supporting Information *
ABSTRACT: The metal−organic frameworks M2(m-dobdc) (M = Mn, Fe, Co, Ni; mdobdc4− = 4,6-dioxido-1,3-benzenedicarboxylate) were evaluated as adsorbents for separating olefins from paraffins. Using single-component and multicomponent equilibrium gas adsorption measurements, we show that the coordinatively unsaturated M2+ sites in these materials lead to superior performance for the physisorptive separation of ethylene from ethane and propylene from propane relative to any known adsorbent, including para-functionalized structural isomers of the type M2(p-dobdc) (p-dobdc4− = 2,5-dioxido-1,4-benzenedicarboxylate). Notably, the M2(m-dobdc) frameworks all exhibit an increased affinity for olefins over paraffins relative to their corresponding structural isomers, with the Fe, Co, and Ni variants showing more than double the selectivity. Among these frameworks, Fe2(m-dobdc) displays the highest ethylene/ethane (>25) and propylene/propane (>55) selectivity under relevant conditions, together with olefin capacities exceeding 7 mmol/g. Differential enthalpy calculations in conjunction with structural characterization of ethylene binding in Co2(m-dobdc) and Co2(p-dobdc) via in situ single-crystal X-ray diffraction reveal that the vast improvement in selectivity arises from enhanced metal−olefin interactions induced by increased charge density at the metal site. Moderate olefin binding enthalpies, below 55 and 70 kJ/mol for ethylene and propylene, respectively, indicate that these adsorbents maintain sufficient reversibility under mild regeneration conditions. Additionally, transient adsorption experiments show fast kinetics, with more than 90% of ethylene adsorption occurring within 30 s after dosing. Breakthrough measurements further indicate that Co2(mdobdc) can produce high purity olefins without a temperature swing, an important test of process applicability. The excellent olefin/paraffin selectivity, high olefin capacity, rapid adsorption kinetics, and low raw materials cost make the M2(m-dobdc) frameworks the materials of choice for adsorptive olefin/paraffin separations.
■
via size-selective,6−10 chemisorptive,11−17 or physisorptive18−22 mechanisms, better performing materials are still needed. Metal−organic frameworks are a class of porous crystalline materials with a high degree of structural tunability that have been demonstrated to be capable of facilitating gas separations through each of these three mechanisms.23−26 Due to the small and similar kinetic diameters of light olefins and paraffins, sizeselective adsorbents typically display moderate selectivities alongside a very low working capacity and slow kinetics. These characteristics arise from the narrow pore sizes necessary to discriminate between small molecules. For example, NbOFFIVE-Ni, a metal−organic framework in which Ni2+-pyrazine square grids are pillared by [NbOF5]2− units to form 4.752(1) Å channels that accommodate propylene but reject propane, displays near perfect propylene/propane selectivity but is
INTRODUCTION
Olefins, including ethylene and propylene, are high-value products obtained primarily from naphtha or ethane cracking and are ubiquitous feedstocks for the most commonly used polymers.1 However, olefins are produced as a mixture with their saturated paraffinic counterparts. Separations of these olefin/paraffin mixtures are some of the most energy-intensive industrial processes and are currently dominated by cryogenic distillation technologies.1,2 The U.S. alone consumes over 120 TBtu/yr in carrying out olefin/paraffin separations.3,4 Nonthermally driven processes, such as adsorption, can dramatically reduce the cost and energy required to purify olefins.5 However, replacing distillation requires adsorbents with adequate performance characteristics, including selectivity, capacity, kinetics, and cost. While there have been significant research efforts directed toward designing materials with the requisite olefin/paraffin separation properties, usually operating © 2017 American Chemical Society
Received: June 20, 2017 Published: October 5, 2017 15363
DOI: 10.1021/jacs.7b06397 J. Am. Chem. Soc. 2017, 139, 15363−15370
Article
Journal of the American Chemical Society impaired by a very low (0.6 mmol/g) working capacity.10 Chemisorptive mechanisms, such as π-complexation with Ag(I) or Cu(I), appear promising due to the high binding enthalpy for olefins.27 However, these high selectivities arise from metal−olefin interactions that are typically greater than 100 kJ/ mol in strength, leading to irreversible binding under typical temperature swing or pressure swing adsorption conditions. By exchanging Ag(I) into the porous aromatic framework PAF-1SO3H,28 Li and co-workers showed high ethylene/ethane selectivity using the 106 kJ/mol binding affinity between the olefin and Ag(I); however, they could not demonstrate reversibility under process conditions.12 Finally, adsorbents that display separation properties based on physisorptive mechanisms typically have faster cycling kinetics and better working capacities due to larger pore sizes and weaker binding affinities. However, due to the difficulty in discriminating between olefins and paraffins, adsorbents have not yet displayed sufficient selectivity to produce polymer grade (99.9% purity) olefins.29 The metal−organic frameworks M2(p-dobdc) (M-MOF-74; CPO-27-M; M = Mg, Mn, Fe, Co, Ni, Zn; p-dobdc4− = 2,5dioxido-1,4-benzenedicarboxylate),30−32 which feature ∼12 Åwide hexagonal channels lined with a high concentration of exposed divalent cations, use coordinatively unsaturated M2+ sites to polarize and adsorb olefins preferentially over paraffins.18,19,33 Among these materials, Fe2(p-dobdc) shows an ethylene/ethane selectivity of ∼14 at 45 °C with an ethylene capacity of greater than 7 mmol/g, and Mn2(p-dobdc) shows a propylene/propane selectivity of ∼16 with a propylene capacity of greater than 7.5 mmol/g. While these frameworks show reversible olefin adsorption with olefin capacities that are more than an order of magnitude higher than in size-selective adsorbents, improvements in selectivity are desired in order to boost olefin purity in the product stream. Such increases in selectivity would translate to olefin purities sufficient for downstream processes, such as polymerization. We hypothesized that higher selectivities could be achieved in these physisorptive materials by altering the affinity of the metal site for adsorbed hydrocarbons. By employing a metasubstituted H4(m-dobdc) ligand, a structural isomer M2(mdobdc) (M = Mg, Mn, Fe, Co, Ni; m-dobdc4− = 4,6-dioxido1,3-benzenedicarboxylate) can be formed. This metal−organic framework has been shown to have increased charge density at the metal sites, leading to enhanced H2 binding enthalpies and greater H2 storage capacities.34 Further, this metal−organic framework is produced from low-cost raw materials, as its linker is derived from a reaction of CO2 with the commodity chemical resorcinol, lending itself to large-scale industrial applications.35 The present study aims to evaluate a series of M2(m-dobdc) metal−organic frameworks for utility in olefin/paraffin separations using single-component equilibrium gas adsorption, multicomponent equilibrium gas adsorption, adsorption kinetics, transient breakthrough measurements, and in situ single-crystal X-ray diffraction experiments. From this, we have found that the M2(m-dobdc) frameworks exhibit superior performance to their para-functionalized counterparts (M2(pdobdc)) and the highest selectivity values among materials that utilize a fast, reversible, physisorptive mechanism. Most notably, Fe2(m-dobdc) shows an ethylene/ethane selectivity of ∼25 and a propylene/propane selectivity of ∼55 under relevant conditions, demonstrating that control over the electronic properties of the open metal sites can lead to improved performance. This is a generalizable concept, in that tuning the
electronic environment around a given adsorption site in a given structure can greatly affect adsorption and separation properties. The combined features of these adsorbents including selectivity, capacity, kinetics, and cost, make the M2(m-dobdc) compounds promising adsorbents for industrial olefin/paraffin separations, and these materials have the potential to offset significant energy consumption relative to the decades-old distillation technology that is employed today.
■
MATERIALS AND METHODS
Synthesis of M2(m-dobdc) (M = Mn, Fe, Co, Ni). The M2(mdobdc) materials were synthesized according to modified versions of the large-scale literature procedures.33 MnCl2, FeCl2, CoCl2, and NiCl2 were purchased from Sigma-Aldrich and used as received. Methanol was purchased from EMD Millipore Corporation as DriSolv grade, dried over 3 Å sieves, and sparged with Ar prior to use. Dimethylformamide (DMF) was purchased from EMD Millipore Corporation as OmniSolv grade, sparged with Ar, and dried with an alumina column prior to use. Co2(m-dobdc) and Ni2(m-dobdc). A mixture of 310 mL of methanol and 310 mL of DMF was added to a 1-L three-neck roundbottom flask equipped with a reflux condenser and purged with N2 while stirring for 1 h. The ligand H4(m-dobdc) (2.00 g, 10.1 mmol) and CoCl2 (3.27 g, 25.2 mmol) or NiCl2 (3.27 g, 25.2 mmol) were added to the solvent under N2 pressure, and the reaction mixture was heated at 120 °C for 18 h while stirring vigorously. The mixture was cooled to ambient temperature and then filtered, yielding a microcrystalline powder. The resulting powder was rinsed with DMF, and soaked in 200 mL of DMF at 120 °C for 24 h. The powder was collected by filtration, rinsed with methanol, and soaked in 200 mL of methanol at 60 °C for 12 h. The supernatant solution was decanted, and 200 mL of fresh methanol were added. This procedure was repeated four times, such that the total time washing with methanol was 2 days. This resulted in an ∼54% yield of Co2(m-dobdc) and Ni2(m-dobdc). The resulting powder was collected by filtration and heated at 180 °C under dynamic vacuum (
